matchmaking-v2: heartbeat progress during the long search

Emit a rotating progress message every ~5s while the Apify scrape + Opus curation run,
so the long (30-40s) silent gap can't idle the response relay and improves the wait UX.

app/agent/session.py

@@ -24,6 +24,23 @@ class MatchmakingAgentSession:
     async def push(self, msg_type: str, **kwargs):
         await self.ws.send_json({"type": msg_type, **kwargs})
 
+    async def _with_heartbeat(self, awaitable, messages: list[str], interval: float = 5.0):
+        """Await `awaitable` while emitting a rotating progress message every `interval`s.
+
+        A live search runs ~30-40s (Apify scrape + Opus curation). Without a steady trickle of
+        messages, the Redis pub/sub channel that relays the result back to the browser idles out and
+        the final `search_complete` is dropped — which is why users were seeing 0 matches. The
+        heartbeat keeps that channel warm AND gives the user real progress during the wait.
+        """
+        task = asyncio.ensure_future(awaitable)
+        i = 0
+        while True:
+            done, _ = await asyncio.wait({task}, timeout=interval)
+            if task in done:
+                return task.result()
+            await self.push("agent_thinking", message=messages[i % len(messages)])
+            i += 1
+
     async def _stub(self, action: str, note: str):
         await self.push("agent_data", action=action, data={"status": "not_implemented", "note": note})
 
@@ -102,7 +119,11 @@ class MatchmakingAgentSession:
         sig = _repo.search_signature(prefs)
         cursors = await _repo.get_search_cursors(self.user_id, sig)
         prefs["_cursors"] = cursors
-        sweep = await search.run_sweep(prefs, fresh=fresh)
+        # Heartbeat through the long Apify scrape so the relay channel stays alive (see _with_heartbeat).
+        sweep = await self._with_heartbeat(
+            search.run_sweep(prefs, fresh=fresh),
+            ["Scanning live roles across the boards…", "Pulling the latest Naukri + LinkedIn + Foundit roles…",
+             "Gathering fresh postings for you…"])
         # Safety net: drop anything already shown / applied / dismissed (covers boards that can't
         # paginate, so the same posting never reappears even if a board re-returns it).
         seen = await _repo.get_seen_ids(self.user_id)
@@ -117,8 +138,11 @@ class MatchmakingAgentSession:
         # Run the blocking LLM work (embeddings + Opus) OFF the event loop, or the long sync OpenAI
         # call freezes the loop and the Redis response can't publish (→ the loader hangs forever).
         top, dbg = await asyncio.to_thread(_sift.sift, prefs, user_context, candidates)
-        await self.push("agent_thinking", message="Scout is reading your top roles…")
+        # Heartbeat through the long Opus curation (the other 30-40s gap that was dropping responses).
-        curated = await asyncio.to_thread(_curate.curate, prefs, user_context, top)   # Opus's call is final
+        curated = await self._with_heartbeat(
+            asyncio.to_thread(_curate.curate, prefs, user_context, top),   # Opus's call is final
+            ["Scout is reading your top roles…", "Scoring fit + writing your report cards…",
+             "Ranking your strongest matches…", "Almost there…"])
         if curated is None:                                       # safety net: sift + templated cards
             curated = _rank.select(top, prefs.get("stretch", "balanced"))
             engine = f"fallback:{dbg['mode']}"